A method and system for assessing the risk of human-wildlife conflict in a nature reserve

Abstract

The present application relates to the technical field of ecological risk assessment, and more particularly to a natural reserve human-wildlife conflict risk assessment method and system, which integrates multi-source data to construct a conflict species list and a standardized distribution data set, constructs a stacked species distribution model based on a maximum entropy model to calculate species hazard degree, matches cultivated land, grassland and population density as exposure indicators for three types of conflicts, namely crop damage, livestock predation and personal injury, constructs a type-specific risk sub-model according to the "hazard degree x exposure degree" framework, superimposes to generate a comprehensive risk raster map, and finally calculates the overall risk value in units of natural reserves. The present application can solve the problem that the existing human-wildlife conflict risk assessment cannot better integrate the contributions of multiple species, distinguish conflict types and adapt to large-scale reserve systems.

Description

Technical Field

[0001] This invention relates to the field of ecological risk assessment technology, and in particular to a method and system for assessing human-wildlife conflict risks in nature reserves. Background Technology

[0002] As the global system of protected areas continues to advance and wildlife populations gradually recover, the overlap between their activity ranges and human living and production spaces is constantly increasing, leading to frequent human-wildlife conflicts. This has become a core issue restricting the effectiveness of nature conservation and affecting community safety, social harmony, and stability. Accurate and scientific spatial identification and assessment of human-wildlife conflict risks are the prerequisites for formulating differentiated prevention measures, optimizing resource allocation, and achieving a balance between conservation and development.

[0003] Currently, research on spatial identification of human-wildlife conflict risks in nature reserves mainly follows two technical paths: one is based on predictive modeling of species and environment, which uses historical conflict sites or species existence records combined with environmental factors such as climate, topography, and land use as inputs, and employs algorithms such as maximum entropy model (MaxEnt), generalized linear model / generalized additive model (GLM / GAM), and random forest to predict potential activity hotspots of single or a few flagship culprit species; the other is based on spatial overlay analysis of multi-dimensional indicators, which constructs an indicator evaluation system that includes dimensions such as species distribution, human activities, and topography, and generates a weighted comprehensive risk index to achieve a spatial presentation of conflict risks.

[0004] However, existing technologies still have significant shortcomings in adapting to the needs of systematic management of large-scale nature reserves: 1. Insufficient scale adaptability: Most methods focus on case studies of single protected areas or local areas, lacking unified evaluation standards and procedures. They cannot support consistent risk identification and cross-protected area prioritization of protected area networks at the national, regional, or even global scales, and are difficult to guide the overall resource allocation across the entire region. 2. Incomplete species coverage: Existing studies mostly include only one or a few flagship species such as giant pandas and tigers, and have not established a standardized method for systematically integrating the cumulative harm of multiple conflict species. This seriously underestimates the overall conflict risk in areas where multiple species coexist, resulting in significant biases in the assessment results. 3. Homogeneous conflict types: The three core conflict types of personal injury, livestock predation, and crop damage were not differentiated in the modeling. Conflicts with different loss forms and occurrence mechanisms were calculated together, resulting in insufficient guidance for specific governance measures in the assessment results and failing to directly guide targeted prevention and management work.

[0005] Therefore, developing a comprehensive assessment framework that can systematically integrate contributions from multiple species, differentiate conflict types, and is applicable to large-scale protected area systems has become a key requirement for the development of this field. Summary of the Invention

[0006] The technical problem solved by this invention is to provide a method and system for assessing human-wildlife conflict risks in nature reserves, so as to solve the problems in the existing technology that human-wildlife conflict risk assessment cannot effectively integrate the contributions of multiple species, distinguish conflict types, or adapt to large-scale protected area systems.

[0007] The basic solution provided by this invention is a method for assessing human-wildlife conflict risks in nature reserves, comprising: S1: Collect species information and environmental factors with conflict behavior records throughout the entire region, and generate a standardized species distribution dataset after preprocessing the recorded species information. S2: Construct a stacked species distribution model based on the maximum entropy model, and calculate the exposure index and hazard index of the target area according to the standardized species distribution dataset; S3: Based on the preset hazard level × exposure level framework, construct sub-models for personal injury risk assessment, livestock predation risk assessment, and crop damage risk assessment respectively. Then, perform spatial overlay calculations on the hazard level grid of each species and the corresponding exposure level grid to obtain a single risk grid of this type. After normalizing each risk grid, overlay them to generate a comprehensive risk grid map of human-wildlife conflict. S4: Using the management boundary of a nature reserve as the statistical unit, extract and calculate the statistical value of the comprehensive risk index of all spatial locations within the statistical unit, which is represented as the overall risk assessment value of the statistical unit.

[0008] Furthermore, S1 includes: S1-1: Construct an initial candidate species database based on authoritative species lists for the target region; S1-2: Extract species information with conflict behavior records from animal diversity databases, verify the initial species, retain conflict species supported by literature or observation records, and populate the species candidate database. S1-3: Search various academic literature databases and mainstream news platforms to select effective cases with clear geographical locations, conflicting species and conflict types, and add them to the species candidate database; S1-4: Cross-integrate the results of S1-2 and S1-3, remove duplicate species, determine the list of species involved in human-wildlife conflict in the study area, and obtain the geographic coordinates of the species in the list based on data from global biodiversity information agencies to obtain a standardized species distribution dataset. S1-5: Obtain climate factors, topographic factors, land use types, and normalized vegetation index of the target area as environmental factors.

[0009] Furthermore, S2 includes: S2-1: Construct a stacked species distribution model based on the maximum entropy model. Take environmental factors and standardized species distribution datasets as inputs to the stacked species distribution model, predict the habitat suitability distribution of the target species, and output the suitability value. S2-2: Divide the target area into multiple pixels. Within each pixel, use the maximum sensitivity plus specificity threshold method to binarize the suitability value, generating the hazard level of the target species, expressed as:

[0010] in, For the s-th species, the pixels in the target region Binarization of hazard level at the location, The s-th species predicted by the stacked species distribution model in pixels The suitability value of continuous habitat at the location; The optimal threshold for maximizing sensitivity and specificity is calculated.

[0011] Furthermore, S2 also includes: S2-3: Divide the target area into crop damage conflict types, livestock predation conflict types, and personal injury conflict types; S2-4: Using cultivated land density to characterize crop damage exposure The expression is:

[0012] in, Represents a pixel The area of ​​cultivated land within the territory, Represents a pixel The total area inside; Grassland density characterizes livestock predation exposure The expression is:

[0013] in, Represents a pixel The area of ​​grassland within; Using population density to characterize personal injury exposure The expression is:

[0014] in, Represents a pixel The number of permanent residents in the area.

[0015] Furthermore, S3 includes: S3-1: Based on the pre-defined hazard × exposure framework, construct sub-models for personal injury risk assessment, livestock predation risk assessment, and crop damage risk assessment, respectively. The expressions are as follows:

[0016] in, Represents a pixel The single risk value grid for the nth type of human-animal conflict, where n represents the conflict type number, n=1 corresponds to crop destruction, n=2 corresponds to livestock predation, and n=3 corresponds to personal injury; This represents the total number of species corresponding to the nth type of conflict; Indicates the s-th species in the n-th type of conflict at pixel level. Binarized hazard level at the location; Represents the cell corresponding to the nth type of conflict. Exposure after minimum-maximum normalization; S3-2: The min-max normalization method is used to map the individual risk value grids of the personal injury risk assessment sub-model, the livestock predation risk assessment sub-model, and the crop damage risk assessment sub-model to... Intervals, after normalization and equal weighting, are superimposed to obtain a pixel-scale comprehensive risk index for human-wildlife conflict, representing a raster map of comprehensive human-wildlife conflict risk. The expression is:

[0017] in, Represents a pixel The overall risk value at the location, Represents a pixel The single risk value grid for Category 1 human-wildlife conflict at the location, Represents a pixel The single risk value grid for Category 2 human-wildlife conflict at the location, Represents a pixel The single risk value grid for Category 3 human-animal conflict.

[0018] Furthermore, the calculation expression for the comprehensive risk index in S4 is as follows:

[0019] in, The overall risk assessment value for the target area. Let be the comprehensive risk value of the j-th pixel within the target area, and m be the total number of pixels within the target area.

[0020] A human-wildlife conflict risk assessment system for nature reserves, applied to the aforementioned human-wildlife conflict risk assessment method for nature reserves, includes: Species information collection and processing module: used to collect species information with conflict behavior records across the entire domain, and generate a standardized species distribution dataset after preprocessing; Environmental Factor Acquisition Module: Used to acquire environmental factors within the target area, including climate factors, topographic factors, land use type, and normalized vegetation index; Indicator Calculation Module: Constructs a stacked species distribution model based on the maximum entropy model, and calculates the exposure index and hazard index of the target area according to the standardized species distribution dataset; Risk assessment sub-model construction module: Based on the preset hazard level × exposure level framework, construct sub-models for personal injury risk assessment, livestock predation risk assessment, and crop damage risk assessment respectively; Human-wildlife conflict comprehensive risk calculation module: It is used to perform spatial superposition calculation of the hazard degree grid of each species and the corresponding exposure degree grid to obtain a single risk grid of this type. After normalizing each risk grid, it is superimposed to generate the comprehensive risk value of human-wildlife conflict. Overall Risk Assessment Calculation Module: This module is used to extract and calculate the statistical value of the comprehensive risk index of all spatial locations within a natural protected area, taking the management boundary of the natural protected area as the statistical unit. This value is then represented as the overall risk assessment value of the statistical unit.

[0021] The technical principle of this invention lies in the following: Based on the core framework of risk science, namely "hazard level × exposure level," this invention integrates a stacked species distribution model with spatial risk grid calculation to construct a risk assessment principle for human-wildlife conflict applicable to large-scale nature reserve systems, as detailed below: First, we integrate global animal diversity databases, academic literature, news cases, and species location data. After preprocessing, we form a conflict animal list and standardized distribution dataset covering the entire region, which solves the problem of incomplete assessment of single species. Subsequently, using climate, topography, land use, and normalized vegetation index as environmental variables, the maximum entropy model was used to simulate the habitat suitability of each conflict species. The species binarized hazard raster was obtained by binarization using the MaxSSS thresholding method, which accurately characterizes the potential activity intensity and spatial distribution of species. Next, exposure indicators are matched according to the conflict occurrence mechanism: farmland density for crop damage, grassland density for livestock predation, and population density for personal injury and death, so as to achieve accurate quantification of the exposure degree of the three types of conflicts. Risk sub-models for the three types of conflicts are constructed respectively. The cumulative harm of multiple species of the same type is multiplied by the normalized exposure space to obtain a single-type risk grid. After normalization, the grid is superimposed with equal weight to generate a pixel-scale comprehensive risk grid. Finally, using the boundaries of nature reserves as statistical units, the average comprehensive risk index within each unit is calculated, transforming spatial grid risk into an overall risk assessment value that can be directly used by the management unit, thus supporting risk ranking and management decisions across protected areas.

[0022] The beneficial effects are: 1. Addressing the issue of insufficient scale adaptability and supporting large-scale systematic management: This invention adopts a unified assessment process and standardized data processing, enabling consistent risk identification at the regional, national, and even global scales. It achieves risk comparison and priority ranking across nature reserves, providing a scientific basis for the overall allocation of conservation resources across the entire region, and breaking through the limitations of traditional methods that are only applicable to a single protected area / local region.

[0023] 2. Addressing the issue of incomplete species coverage and accurately quantifying the cumulative risk of multiple species: By integrating the hazard levels of multiple conflicting species through a stacked species distribution model system, the spatial superposition calculation of the hazard of multiple species is realized, avoiding the underestimation of risk caused by assessing only a single or a few species, and significantly improving the accuracy and completeness of risk assessment in multi-species coexistence areas.

[0024] 3. Address the issue of homogeneous conflict types and enhance the targeting of governance: Independent sub-models are constructed for the three core conflict types of personal injury, livestock predation, and crop damage. Exposure indicators are matched differently to distinguish the occurrence mechanism and loss form of different conflicts. The assessment results can directly guide the formulation of targeted prevention measures, which solves the defects of traditional methods such as mixed assessment and weak governance targeting.

[0025] 4. Achieve precise grid-based assessment and statistical coupling of management units: It takes into account both the precise spatial distribution at the pixel scale and the overall risk assessment of nature reserve management units. It not only locates high-risk spatial points, but also provides the overall risk level of the protected area, and outputs risk maps and quantitative evaluation values, adapting to the daily management and decision-making needs of nature reserves.

[0026] 5. Diverse data sources, standardized methods, and strong scalability: Based on publicly available databases and standardized model algorithms, it does not rely on local proprietary data. The methods and processes are standardized and replicable, applicable to human-wildlife conflict risk assessment in various nature reserves, and have broad application value. Attached Figure Description

[0027] Figure 1 This is a flowchart of an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the implementation steps of an embodiment of the present invention; Figure 3 This is a schematic diagram of the conflict species list screening process according to an embodiment of the present invention; Figure 4 This is a functional block diagram of an embodiment of the present invention. Detailed Implementation

[0028] The following detailed description illustrates the specific implementation method: The implementation examples are basically as follows Figure 1 and Figure 2 As shown: A method for assessing human-wildlife conflict risks in nature reserves, including: S1: Collect species information and environmental factors with conflict behavior records across the entire region. After preprocessing the recorded species information, generate a standardized species distribution dataset. Specifically, all spatial data is uniformly preprocessed to the WGS84 coordinate system with a 1km × 1km spatial resolution, such as... Figure 3 As shown, S1 includes: S1-1: Construct an initial species candidate database based on the authoritative species list of the target area; in this embodiment, the authoritative species list adopts the "List of Mammals in China (2024 Edition)" to construct the initial species candidate database.

[0029] S1-2: Extract species information with conflict behavior records from animal diversity databases, verify the initial species, retain conflict species supported by literature or observation records, and populate the species candidate database; specifically, access the Animal Diversity Web database and perform the following operations: First, using the administrative boundary vector data of the study area as a spatial filtering condition, all terrestrial mammal species with natural distribution records in the area are retrieved to generate a basic list of mammals in the region. Subsequently, the "ecological habits" and "interactions with humans" fields of each species in the basic list were checked one by one, and species that were clearly recorded to have any conflict behavior such as personal injury, livestock predation, or crop damage were screened. Finally, species that are extinct, only in captivity, or have no wild population records were removed, forming a preliminary candidate database of species for human-wildlife conflict.

[0030] like Figure 3 As shown, 40 conflict species were identified through screening.

[0031] S1-3: Search various academic literature databases and mainstream news platforms to select valid cases with clear geographical locations, conflicting species, and conflict types, and add them to the species candidate database; specifically, conduct a systematic search and screening of academic literature and public news reports from the past 20 years: 1. Academic literature retrieval: Search CNKI, Web of Science Core Collection, and ScienceDirect database. The search formula is: (human-wildlife conflict OR wildlife harm OR wildlife conflict) AND (nature reserve OR protected area) AND (personal injury OR livestock predation OR crop damage) AND research area name; 2. News report retrieval: Search mainstream news platforms with the keywords: [Study area] wild animals injuring people / eating livestock / destroying crops; 3. Screening of Valid Cases: Cases that meet the following conditions will be retained: a. The specific geographical location of the conflict is clearly recorded (accuracy no less than the county level); b. The official Chinese name or Latin name of the species involved is clearly recorded; c. The type of conflict is clearly recorded (personal injury / livestock predation / crop damage). 4. Add species that appeared in the valid cases but were not included in the preliminary candidate pool to the candidate pool, and label the main conflict type corresponding to each species.

[0032] like Figure 3 As shown, this embodiment identified a total of 39 conflict species.

[0033] S1-4: By cross-integrating the results of S1-2 and S1-3, duplicate species were removed, a list of species involved in human-wildlife conflict in the study area was determined, and geographic coordinates of the occurrence points of species in the list were obtained from global biodiversity information agencies to obtain a standardized species distribution dataset; specifically: First, the candidate species obtained from S1-2 and S1-3 were cross-compared to eliminate duplicate species, synonyms, and misidentified species, and finally a list of target species for human-animal conflict in the study area was formed, and the main conflict type corresponding to each species in the list was identified. Next, based on the Global Biodiversity Information Institute (GBIF), the occurrence location data of all species in the list were downloaded in batches. The data fields include the species' Latin name, geographical coordinates (longitude / latitude), record time, and data source. The data points were then processed, including deduplication: deleting duplicate records of the same species in the same pixel; spatial verification: overlaying all species locations with the official distribution range vector data of the corresponding species published by the International Union for Conservation of Nature (IUCN), and deleting outlier locations that are outside the IUCN distribution range; and precision filtering: deleting low-quality locations with coordinate precision below 1km. Finally, the field survey data on species occurrence locations and scientific monitoring data provided by the management departments of nature reserves within the study area were integrated and added to the cleaned location data to generate a standardized species distribution dataset that can be used for model training.

[0034] S1-5: Obtain climate factors, topographic factors, land use types, and normalized vegetation index of the target area as environmental factors; specifically, the acquisition paths for each environmental factor are as follows: Climate factors: Nineteen bioclimatic variables were obtained from the WorldClim database, including annual average temperature, annual precipitation, annual temperature range, minimum temperature of the coldest month, maximum temperature of the hottest month, and seasonal variation coefficient of precipitation, with a spatial resolution of 1 km. Topographic factors: Digital elevation model (DEM) data were acquired from the Space Shuttle Radar Topography Mission (SRTM) with a spatial resolution of 30m. After resampling to 1km, three topographic factors, namely elevation, slope, and aspect, were extracted using GIS spatial analysis tools. Land use types: Global land cover data (such as ESA CCI-LC, MODIS MCD12Q1) for the corresponding year of the study area were obtained and reclassified into six categories: cultivated land, forest land, grassland, construction land, water area, and unused land, with a spatial resolution of 1 km. Normalized Difference Vegetation Index (NDVI): Monthly composite MODIS NDVI data for the corresponding year in the study area were obtained, and the annual average NDVI value was calculated with a spatial resolution of 1 km. Finally, a unified projection transformation and resampling were performed on all environmental factors to ensure that the coordinate system, spatial resolution, and spatial range of all data were completely consistent, thus generating a standardized environmental factor dataset.

[0035] S2: Construct a stacked species distribution model based on the maximum entropy model, and calculate the exposure index and hazard index of the target area according to the standardized species distribution dataset; wherein, S2 includes: S2-1: Construct a stacked species distribution model based on the maximum entropy model. Environmental factors and a standardized species distribution dataset are used as inputs to the stacked species distribution model to predict the habitat suitability distribution of the target species and output suitability values. Specifically: First, the four environmental factors obtained from S1-4—climate, topography, land use, and NDVI—were tested for collinearity. The variance inflation factor (VIF) method was used to remove highly collinear factors with VIF values ​​greater than 10, and independent environmental variables were retained as model inputs to avoid model overfitting caused by multicollinearity. Next, MaxEnt version 3.4.4 was used to build the model, with the training set accounting for 75% and the test set accounting for 25%. The regularization coefficient was set to 1, the model was run 10 times, and the output type was selected as "logistic". The model obtained continuous habitat suitability values ​​in the range of 0-1. The larger the value, the higher the habitat suitability of the target species. Subsequently, the standardized species distribution dataset generated in S1-3 and the preprocessed environmental factors were input into the model, and training was completed for each species. The area under the receiver operating characteristic curve (AUC) was used to verify the model accuracy. Valid model results with an AUC value greater than 0.7 were retained, species models with insufficient accuracy were removed, and the parameters were re-optimized. Finally, the habitat suitability prediction results of all valid species are integrated to form a stacked species distribution model output set, which yields a continuous habitat suitability grid for each species in the target area, representing the suitability value.

[0036] S2-2: Divide the target area into several 1km×1km regular pixels. Within each pixel, use the maximum sensitivity plus specificity threshold method to binarize the suitability value to generate the hazard level of the target species. The expression is:

[0037] in, For the s-th species, the pixels in the target region The binarized hazard level is set to 1, indicating that the pixel is a high-suitability area for species activity with conflict hazards; and 0, indicating that the pixel is a low-suitability area for species activity with no conflict hazards. The s-th species predicted by the stacked species distribution model in pixels The suitability value of continuous habitat at the location; The optimal threshold for maximizing sensitivity and specificity is automatically calculated by the MaxEnt model based on training data. This threshold can simultaneously maximize the model's sensitivity and specificity, balancing the errors of missed and false detections of species habitats.

[0038] S2-3: Divide the target area into three conflict types: crop damage conflict, livestock predation conflict, and personal injury conflict; and further divide it into three independent subsets: Subset of species involved in crop damage conflict: includes species whose main damaging behaviors are eating and trampling crops; Subset of species involved in livestock predation conflict: includes species whose primary harmful behavior is predation of domestic livestock; Subset of species involved in personal injury conflicts: includes species whose primary harmful behavior is attacking humans and causing personal injury or death.

[0039] If a species exhibits multiple harmful behaviors, they should be categorized into corresponding subsets based on their primary conflict type to ensure the accuracy of subsequent cumulative harm calculations.

[0040] S2-4: Using cultivated land density to characterize crop damage exposure The expression is:

[0041] in, Represents a pixel The area of ​​cultivated land within the territory, in units It is extracted from the land use type data obtained from S1-4; Represents a pixel Total area within, in units ; Grassland density characterizes livestock predation exposure The expression is:

[0042] in, Represents a pixel Grassland area within, unit It is extracted from the land use type data obtained from S1-4.

[0043] Using population density to characterize personal injury exposure The expression is:

[0044] in, Represents a pixel The number of permanent residents in the area.

[0045] S3: Based on a pre-defined hazard × exposure framework, sub-models for assessing personal injury risk, livestock predation risk, and crop damage risk are constructed. The hazard raster for each species is spatially overlaid with its corresponding exposure raster to obtain a single risk raster of that type. These risk raster rasters are then normalized and overlaid to generate a comprehensive human-wildlife conflict risk raster map. S3 includes: S3-1: Based on the pre-defined hazard × exposure framework, construct sub-models for personal injury risk assessment, livestock predation risk assessment, and crop damage risk assessment, respectively. The expressions are as follows:

[0046] in, Represents a pixel The single risk value grid for the nth type of human-animal conflict, where n represents the conflict type number, n=1 corresponds to crop destruction, n=2 corresponds to livestock predation, and n=3 corresponds to personal injury; This represents the total number of species corresponding to the nth type of conflict; Indicates the s-th species in the n-th type of conflict at pixel level. Binarized hazard level at the location; Represents the cell corresponding to the nth type of conflict. Exposure after minimum-maximum normalization; Perform the above calculations separately to obtain a single risk grid for crop damage. Single risk grid for livestock predation Single risk grid for personal injury The larger the value of each grid cell, the higher the potential risk of this type of conflict at the corresponding location.

[0047] S3-2: The min-max normalization method is used to map the individual risk value grids of the personal injury risk assessment sub-model, the livestock predation risk assessment sub-model, and the crop damage risk assessment sub-model to... Intervals, after normalization and equal weighting, are superimposed to obtain a pixel-scale comprehensive risk index for human-wildlife conflict, representing a raster map of comprehensive human-wildlife conflict risk. The expression is:

[0048] in, Represents a pixel The overall risk value at the location, Represents a pixel The single risk value grid for Category 1 human-wildlife conflict at the location, Represents a pixel The single risk value grid for Category 2 human-wildlife conflict at the location, Represents a pixel The single risk value grid for Category 3 human-animal conflict.

[0049] A comprehensive risk index based on all pixels This generates a comprehensive human-wildlife conflict risk grid map of the entire target area, visually presenting the spatial distribution pattern of conflict risks. High-value areas are potential hotspots for high-incidence conflicts, providing a spatial basis for subsequent risk statistics and precise prevention and control of protected area units.

[0050] S4: Using the management boundary of a nature reserve as the statistical unit, extract and calculate the statistical value of the comprehensive risk index for all spatial locations within the statistical unit. This value is represented as the overall risk assessment value of the statistical unit. The formula for calculating the comprehensive risk index is as follows:

[0051] in, The overall risk assessment value for the target area. Let be the comprehensive risk value of the j-th pixel within the target area, and m be the total number of pixels within the target area.

[0052] Based on the above technical solution, this application embodiment compiles and calculates a risk assessment of human-wildlife conflict in Chinese nature reserves: The risk index statistics show that the average comprehensive risk of China's nature reserves is 0.1299 (std = 0.1270), with approximately 86% of reserves facing at least one type of human-wildlife conflict risk, indicating that human-wildlife conflict has become a common problem faced by most nature reserves. The reserves with the highest overall risk are concentrated in the southwestern mountainous areas bordering Sichuan, Yunnan, and Guizhou provinces, with some scattered high-risk reserves distributed in a few areas in the central and southeastern regions. The average values ​​of the three types of conflict risks differ significantly, with the highest average risk of personal injury (0.0936±0.1001), followed by the risk of livestock predation (0.0215±0.0528), and the lowest average risk of crop damage (0.0156±0.0416). In terms of risk distribution, high-risk reserves for crop damage are mainly located in the low mountains and hills of the southwest, and scattered in some agricultural clusters in the southeast and central China. High-risk reserves for livestock predation are significantly concentrated on the edge of the Qinghai-Tibet Plateau and high-altitude pastoral areas, indicating that the problem of livestock predation within pastoral protected areas and their controlled areas is the most prominent. High-risk areas for personal injury are mainly concentrated in several protected areas in the southwestern mountainous region, and show localized high values ​​in several protected areas in the central and southeastern coastal regions.

[0053] In another embodiment of this example, a human-wildlife conflict risk assessment system for nature reserves is also included, applied to the aforementioned human-wildlife conflict risk assessment method for nature reserves, such as... Figure 4 As shown, it includes: Species information collection and processing module: used to collect species information with conflict behavior records across the entire domain, and generate a standardized species distribution dataset after preprocessing; Environmental Factor Acquisition Module: Used to acquire environmental factors within the target area, including climate factors, topographic factors, land use type, and normalized vegetation index; Indicator Calculation Module: Constructs a stacked species distribution model based on the maximum entropy model, and calculates the exposure index and hazard index of the target area according to the standardized species distribution dataset; Risk assessment sub-model construction module: Based on the preset hazard level × exposure level framework, construct sub-models for personal injury risk assessment, livestock predation risk assessment, and crop damage risk assessment respectively; Human-wildlife conflict comprehensive risk calculation module: It is used to perform spatial superposition calculation of the hazard degree grid of each species and the corresponding exposure degree grid to obtain a single risk grid of this type. After normalizing each risk grid, it is superimposed to generate the comprehensive risk value of human-wildlife conflict. Overall Risk Assessment Calculation Module: This module is used to extract and calculate the statistical value of the comprehensive risk index of all spatial locations within a natural protected area, taking the management boundary of the natural protected area as the statistical unit. This value is then represented as the overall risk assessment value of the statistical unit.

[0054] The above are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for assessing human-wildlife conflict risks in nature reserves, characterized in that: include: S1: Collect species information and environmental factors with conflict behavior records throughout the entire region, and generate a standardized species distribution dataset after preprocessing the recorded species information. S2: Construct a stacked species distribution model based on the maximum entropy model, and calculate the exposure index and hazard index of the target area according to the standardized species distribution dataset; S3: Based on the preset hazard level × exposure level framework, construct sub-models for personal injury risk assessment, livestock predation risk assessment, and crop damage risk assessment respectively. Then, perform spatial overlay calculation on the hazard level grid of each species and the corresponding exposure level grid to obtain a single risk grid of the corresponding sub-model. After normalizing each risk grid, overlay them to generate a comprehensive risk grid map of human-wildlife conflict. S4: Using the management boundary of nature reserves as the statistical unit, extract and calculate the statistical value of the comprehensive risk index of all spatial locations within the statistical unit, which is represented as the overall risk assessment value of the statistical unit; S2 includes: S2-1: Construct a stacked species distribution model based on the maximum entropy model. Take environmental factors and standardized species distribution datasets as inputs to the stacked species distribution model, predict the habitat suitability distribution of the target species, and output the suitability value. S2-2: Divide the target area into multiple pixels. Within each pixel, use the maximum sensitivity plus specificity threshold method to binarize the suitability value, generating the hazard level of the target species, expressed as: in, For the s-th species, the pixels in the target region Binarization of hazard level at the location, The s-th species predicted by the stacked species distribution model in pixels The suitability value of continuous habitat at the location; The optimal threshold for maximum sensitivity and specificity is calculated. S2 further includes: S2-3: Divide the target area into crop damage conflict types, livestock predation conflict types, and personal injury conflict types; S2-4: Using cultivated land density to characterize crop damage exposure The expression is: in, Represents a pixel The area of ​​cultivated land within the territory, Represents a pixel The total area inside; Grassland density characterizes livestock predation exposure The expression is: in, Represents a pixel The area of ​​grassland within; Using population density to characterize personal injury exposure The expression is: in, Represents a pixel The number of permanent residents in the area.

2. The method for assessing human-wildlife conflict risk in nature reserves according to claim 1, characterized in that: S1 includes: S1-1: Construct an initial candidate species database based on authoritative species lists for the target region; S1-2: Extract species information with conflict behavior records from animal diversity databases, verify the initial species, retain conflict species supported by literature or observation records, and populate the species candidate database. S1-3: Search various academic literature databases and mainstream news platforms to select effective cases with clear geographical locations, conflicting species and conflict types, and add them to the species candidate database; S1-4: Cross-integrate the results of S1-2 and S1-3, remove duplicate species, determine the list of species involved in human-wildlife conflict in the study area, and obtain the geographic coordinates of the species in the list based on data from global biodiversity information agencies to obtain a standardized species distribution dataset. S1-5: Obtain climate factors, topographic factors, land use types, and normalized vegetation index of the target area as environmental factors.

3. The method for assessing human-wildlife conflict risk in nature reserves according to claim 2, characterized in that: S3 includes: S3-1: Based on the pre-defined hazard × exposure framework, construct sub-models for personal injury risk assessment, livestock predation risk assessment, and crop damage risk assessment, respectively. The expressions are as follows: in, Represents a pixel The single risk value grid for the nth type of human-animal conflict, where n represents the conflict type number, n=1 corresponds to crop destruction, n=2 corresponds to livestock predation, and n=3 corresponds to personal injury; This represents the total number of species corresponding to the nth type of conflict; Indicates the s-th species in the n-th type of conflict at pixel level. Binarized hazard level at the location; Represents the cell corresponding to the nth type of conflict. Exposure after minimum-maximum normalization; S3-2: The min-max normalization method is used to map the individual risk value grids of the personal injury risk assessment sub-model, the livestock predation risk assessment sub-model, and the crop damage risk assessment sub-model to... Intervals, after normalization and equal weighting, are superimposed to obtain a pixel-scale comprehensive risk index for human-wildlife conflict, representing a raster map of comprehensive human-wildlife conflict risk. The expression is: in, Represents a pixel The overall risk value at the location, Represents a pixel The single risk value grid for Category 1 human-wildlife conflict at the location, Represents a pixel The single risk value grid for Category 2 human-wildlife conflict at the location, Represents a pixel The single risk value grid for Category 3 human-animal conflict.

4. The method for assessing human-wildlife conflict risk in nature reserves according to claim 3, characterized in that: The formula for calculating the comprehensive risk index in S4 is as follows: in, The overall risk assessment value for the target area. Let be the comprehensive risk value of the j-th pixel within the target area, and m be the total number of pixels within the target area.

5. A human-wildlife conflict risk assessment system for nature reserves, applied to the human-wildlife conflict risk assessment method for nature reserves as described in any one of claims 1-4, characterized in that: include: Species information collection and processing module: used to collect species information with conflict behavior records across the entire domain, and generate a standardized species distribution dataset after preprocessing; Environmental Factor Acquisition Module: Used to acquire environmental factors within the target area, including climate factors, topographic factors, land use type, and normalized vegetation index; Indicator Calculation Module: Constructs a stacked species distribution model based on the maximum entropy model, and calculates the exposure index and hazard index of the target area according to the standardized species distribution dataset; Risk assessment sub-model construction module: Based on the preset hazard level × exposure level framework, construct sub-models for personal injury risk assessment, livestock predation risk assessment, and crop damage risk assessment respectively; Human-wildlife conflict comprehensive risk calculation module: It is used to perform spatial superposition calculation of the hazard degree grid of each species and the corresponding exposure degree grid to obtain the single risk grid of the corresponding sub-model. After normalizing each risk grid, it is superimposed to generate the comprehensive risk value of human-wildlife conflict. Overall Risk Assessment Calculation Module: This module is used to extract and calculate the statistical value of the comprehensive risk index of all spatial locations within a natural protected area, using the management boundary of the natural protected area as the statistical unit. This value is then represented as the overall risk assessment value of the statistical unit.

Images